基于自适应安全区域的轨道实时化简方法  

A Safe-Region Based Adaptive Method for Real-Time Trajectory Simplification

作　　者：李文海[1,2] 程志光[1] 文卫东[1] 向隆刚[3] 郭晓倩[1] 

机构地区：[1]武汉大学计算机学院,武汉430072 [2]软件工程国家重点实验室,武汉430072 [3]测绘遥感信息工程国家重点实验室,武汉430079

出　　处：《计算机学报》2014年第9期1922-1935,共14页Chinese Journal of Computers

基　　金：粗监督时序轨道分析若干关键技术研究(60903035);时空观测数据的多尺度聚集可视化分析(41001296);国家"八六三"高技术研究发展计划项目基金(2013AA12A301);湖北省重大科技创新计划项目(2013AAA020)资助~~

摘　　要：利用采样思想在移动端对原始位置序列进行即时处理,移动轨迹实时化简为服务端提供误差可控的简化序列.由于移动规律、运动方向、运动速率等变化因素的不可预知,现有方法的插值还原精度和轨迹化简率易受到运动特征局部变化的影响.在已有工作基础上,利用采样点速率和方向矢量为轨迹建立总体和局部两个层次的误差度量.通过定义安全区域LDAR和TVAR限制插值点的误差范围,将相邻关键点之间的总体误差和相邻轨迹点之间的局部误差的容差范围作为采样监督参数,构造自适应于轨迹方向和速率的实时化简算法.实验结果表明,文中提出的算法对于平稳轨迹能够进行高效的实时化简.同时,对于速率和方向等参数的波动也较现有的推算定位方法有更强的鲁棒性.We are witnessing a proliferation of Internet-worked, geo-positioned mobile devices.It is becoming increasingly feasible to monitor and query data that capture the continuouslyevolving states of the real-life phenomena. Nevertheless, the sampled trajectory series as well astheir reduced expressions are highly influenced by the variant moving characters. This paperpresents a two-layered constraint framework for simplifying the trajectory in real-time. Boundedby two categories of safe regions, the Dead Reckoning guided simplification is divided into twophases. LDAR, a technique for controlling the abnormal position, constructs the local safe regionin terms of the speed and the direction. On this basis, in order to reduce the temporal metricbetween any two consequent reduced points, TVAR further bounds the striped scope in a totalviewpoint. Empirical studies on several real workloads demonstrate that our approaches outper-form the sampling methods by orders of magnitudes, and the adaptive strategy is more efficientand robust than state-of-the-art competitors.

关 键 词：推算定位 安全区域 轨迹化简 移动对象 时空数据库 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]